Estimation of plates impression by restaurant information on the gourmet website

The selection of plates plays a significant role as a means of presentation that enhances a dish's attractiveness. The Quality of Dish (QoD), a fulfilling experience of eating, improves by a selection of plates depending on the theme of the dining area. However, selecting the proper plates for a specific meal can be daunting for ordinary individuals, underscoring the need for a system to simplify this process. The research aims to create a system that supports plate selection. Such a system would recommend plates based on the impression assigned to each plate. However, collecting knowledge to use such a system is problematic. To address the problem, we focused on the restaurant information on the gourmet site. We set up the hypothesis that the view from the restaurant inside and the plates used in the restaurant give the same impression. We got word of mouth and dish pictures from the gourmet site based on the hypothesis. Using comments on restaurant impressions, we created a dataset of plates to which impressions were given. We tried assigning plates the impression by calculating the degree of similarity. In this paper, we examine the validity of measuring similarity between plates based on image features and the degree of agreement between impressions of similar plates. The results suggest that the proposed similarity calculation method has room for further study The impressions given to the plates by the proposed method were also seen in similar plates to a certain degree.